EP0262944B1 - Error correction apparatus - Google Patents
Error correction apparatus Download PDFInfo
- Publication number
- EP0262944B1 EP0262944B1 EP87308648A EP87308648A EP0262944B1 EP 0262944 B1 EP0262944 B1 EP 0262944B1 EP 87308648 A EP87308648 A EP 87308648A EP 87308648 A EP87308648 A EP 87308648A EP 0262944 B1 EP0262944 B1 EP 0262944B1
- Authority
- EP
- European Patent Office
- Prior art keywords
- circuit
- error
- encoding
- correction
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/35—Unequal or adaptive error protection, e.g. by providing a different level of protection according to significance of source information or by adapting the coding according to the change of transmission channel characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/60—Methods or arrangements for performing computations using a digital non-denominational number representation, i.e. number representation without radix; Computing devices using combinations of denominational and non-denominational quantity representations, e.g. using difunction pulse trains, STEELE computers, phase computers
- G06F7/72—Methods or arrangements for performing computations using a digital non-denominational number representation, i.e. number representation without radix; Computing devices using combinations of denominational and non-denominational quantity representations, e.g. using difunction pulse trains, STEELE computers, phase computers using residue arithmetic
- G06F7/724—Finite field arithmetic
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/13—Linear codes
- H03M13/15—Cyclic codes, i.e. cyclic shifts of codewords produce other codewords, e.g. codes defined by a generator polynomial, Bose-Chaudhuri-Hocquenghem [BCH] codes
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/13—Linear codes
- H03M13/15—Cyclic codes, i.e. cyclic shifts of codewords produce other codewords, e.g. codes defined by a generator polynomial, Bose-Chaudhuri-Hocquenghem [BCH] codes
- H03M13/151—Cyclic codes, i.e. cyclic shifts of codewords produce other codewords, e.g. codes defined by a generator polynomial, Bose-Chaudhuri-Hocquenghem [BCH] codes using error location or error correction polynomials
Definitions
- the present invention relates to Galois field arithmetic apparatus, particularly for use in the field of error correction, intended for a reduction of rate of occurrence of error in the path of transmission from a device such as an optical disk device, opto-magnetic disk device, and so forth.
- the invention relates to an error correction circuit for use in a digital electronic device such as a digital audio device, particularly a flag strategy setting circuit which is typically a BCH (Bose-Chaudhuri-Hocquenghem) encoding and decoding circuit for use in the error correction circuit.
- a flag strategy setting circuit which is typically a BCH (Bose-Chaudhuri-Hocquenghem) encoding and decoding circuit for use in the error correction circuit.
- the invention is concerned with a syndrome creating circuit incorporated in the BCH encoding and decoding circuit, particularly to an element-exponent/vector conversion circuit, multiplication circuit and division circuit on a Galois function.
- Galois function in this specification is used to mean a set of a definite number of elements which can be processed by addition, subtraction, multiplication and division and usually expressed as GF(q), where q represents the number of elements.
- error correction apparatus of the type described has been generally designed to operate in response to an error correction code of a predetermined format having factors such as the length of error correction code, correction capacity and interleave.
- the known error correction apparatus could not operate when the correction code of a different format is applied and when a plurality of blocks of correction codes of various formats are supplied serially.
- Optical disks and opto-magnetic disks are usually kept under sector administration.
- the sector contains not only the ordinary data but also address data. Protection of these two types of data is a matter of great significance.
- error correction codes such as read solomon codes have been used as the protection means for protecting the ordinary data portion
- error detection code such as CRC (Cyclic Redundancy Check) codes have been used as means for protecting the address data.
- CRC Cyclic Redundancy Check
- any detection error causes a click noise and, hence, is critical. Any error which cannot be corrected, however, can be compensated for by interpolation to such an extent that it does not substantially impair the audio effect, provided that the error is detected. It is therefore necessary to maximize the error detection capacity at the cost of reduced error correction capacity, as in the simple strategy (1) shown above.
- 2 symbol error correction is performed by the C1 decoder as in the case of the super strategy (2) mentioned above. In this case, flag is set up to enhance also the error detection capacity, thereby to reduce the rate of error detection failure.
- the conventional system employs different circuits for encoding and decoding. Attempts have been made to conduct encoding and decoding by making use of the same circuit in a multiplied manner in accordance with, for example, microprogramming. In such a case, however, it is necessary that delicate difference be made between the encoding process and the decoding process. Therefore, when the encoding and decoding are to be conducted by means of the same circuit board or chip, the number of circuits or the capacity of ROM is increased inevitably. This in turn causes the size of the device such as an optical disk device to be increased undesirably.
- the syndrome creating circuit in conventional BHC encoding or decoding circuit produces the syndrome S as represented by the following formula (2), upon receipt of a word J which, as shown in the following formula (1), has a term of the code word I and a term of error E.
- the syndrome S is created as the product of a check matrix H and the received word J, as shown by the formula (2).
- n represents the code length
- H represents the check matrix
- J represents the code length
- S represents the syndrome
- I represents the code word
- E represents the error
- the syndrome creating circuit can be expressed as follows for each syndrome Si.
- the syndrome creating circuit on the Galois function GF(2q) can be constructed as shown in Fig. 3.
- the syndrome creating circuit for realizing SO to S3 expressed by the formula (2) can be constructed as shown in Fig. 5.
- CK represents clock for each received word Ji
- CL represents the clear for each code length.
- the size of the construction shown in Fig. 5 becomes large as the number q or the number i becomes large.
- the arrangement shown in Fig. 5 is not suitable for use in the case where the syndromes SO to S3 are interconnected and processed through a BUS line.
- the VE (vector to exponent) conversion and EV (exponent to vector) conversion can be conducted by making use of a ROM.
- EP-A-0080528 discloses a Galois field multiplication apparatus implemented with a logic circuit rather than ROM.
- the present invention provides multiplication apparatus (Fig 55) for multiplying two arbitrary elements X and Y on a Galois Field GF(2 m ), comprising;
- the invention provides a method of digital signal processing which includes the step of multiplying together two arbitrary elements X, Y on a Galois Field GF(2 m ) which comprises the steps of;
- a read solomon code (referred to as "RS code” hereinafter) as an example of error correction signal typically has a cross-interleaved construction as shown in Fig. 10. This means that each block requires two times of error correction as represented by C1 and C2. When input data is supplied without any interval, the processing delay time will be accumulated unless an erasure correction and error correction unit is provided for each of C1 and C2. order to simplify the construction of the hardware, therefore, this embodiment is constructed such that a single erasure correction and error correction unit is made to operate at a doubled speed as denoted by 10 and 02 in Fig.
- the erasure correction and error correction unit operates at a normal speed in the when there is no interleave or when the data transmission speed is high.
- data is input as represented by data-in in response to gate-in and, after a computation for error correction is completed, the corrected data is output as represented by data-out.
- the data input/output time in the doubled speed mode 1 and 2 is 1/2 that of the normal input/output time.
- the erasure correction and error. correction unit has, as shown in Fig. 12, a RAM 121 for storing the input data, an erasure correct and error correct block 122 and a control block 123 for varying the timing signal for driving the block 122 in accordance with an input format select signal.
- this embodiment makes use of an algorithm as shown in Fig. 13, which can simplify the hardware construction and, hence, facilitates the operation under the condition where the timing signal is variable.
- Fig. 14 shows the construction of decoding hardware adopted when this algorithm is used.
- the encoding is conducted by dividing data by a creating polynominant. In this embodiment, however, the encoding is conducted by varying the control of the error correction unit which is used in decoding.
- the construction of a hardware used for this purpose is shown in Fig. 15.
- Fig. 16 shows the flow of the process for creating parity bit.
- the decoding and encoding described above can be united with each other by a hardware as shown in Fig. 17.
- the factor A -1 can be definitely determined provided that i, j, k and l are given.
- Fig. 18 is a block diagram of the erasure correction and error correction circuit. Since i, j, k and I are definite, the A- 1 creating circuit can be determined by several ROMs. Therefore, the encoding/decoding circuit including the erasure correction function can be constructed as shown in Fig. 1.
- the correcting operation is delayed behind the generation of syndrome by a period corresponding to one received word.
- the received word therefore, has to be temporarily stored in buffer 13.
- the error pattern computed at the time of generation of syndrome has to be temporarily stored in the addresses i, j, k and I of a buffer 54, in order that the pattern is output at the positions i, j, k and I in synchronization with the output from the buffer 13.
- i, j, k and I is output from the AD, this value is output so that the error pattern is output in synchronization with the output from the buffer 13.
- this device When there is no interleave in the error correction code or when the processing is conducted at the doubled speed, this device is used only for a single error correction. In such a case, this device can be used as it is without any modification.
- this device is used for two times of error correcting operation.
- error correction is conducted both in the vertical and horizontal directions as shown by arrows in Fig. 10.
- the operation of the device at the doubled speed can be conducted by doubling the rates of clock and trigger. In this case, however, a difficulty is encountered due to difference in the format between C1 and C2.
- the format can be varied by imparting different values of the correction capacity T and code lengths n, k to C1 and C2.
- Fig. 19 shows a circuit which is adopted when the error correction code has an interleaved structure.
- the code length k can be determined by n-2. T-E. It is thus possible to conduct the desired interleaved processing.
- an error correction apparatus can be driven for general purposes.
- this apparatus in one chip and designing the same to be adaptable to a variety of formats, it becomes possible to apply the invention not only to optical disk and opto-magnetic disk devices but also to all types of transmission path which suffers from large rate of occurrence of error.
- a mark S.M. represents a sector mark which indicates the beginning end of a sector.
- ID is read in accordance with clock component extracted by SYNC.
- a symbol A.M. in ID represents an address mark which indicates the beginning of an address.
- GAP is formed between ID section and DATA section for the purpose of absorbing jitter, thereby to accommodate any offset of ID section from data section including SYNC, S.M. and DATA. The same applies also to the data section. The last GAP is preserved for the purpose of absorbing any offset between the present sector and the next sector.
- This data structure can be processed by the apparatus of the present invention, even though the address mark and the data section are coded in different formats.
- an error correction apparatus having a general-purpose format selector section and an erasure correction selecting section including erasure position input section is constructed in a single chip so that a single erasure correction and error correction apparatus can operate for a variety of formats.
- error correction can be conducted by a single erasure correction and error correction block by employing a select signal for selecting different formats.
- a control unit may be employed which enables a single erasure correction and error correction block to operate at a doubled speed when the data to be corrected has an interleaved structure and at a normal speed when the error correction code does not have any interleaved structure.
- the apparatus of the described embodiment when applied to erasure correction and error correction of data having a sector construction, the error correction can be conducted not only on the data in the sector but also on address data of the sector by means of the same erasure correction and error correction block.
- a circuit as shown in Fig. 24 is constructed by using L1 and L2 of an algorithm which follows formulae (12) to (21) shown below.
- the timing of operation of this circuit arrangement is shown in Fig. 25.
- n represents the code length
- H represents the check matrix
- J represents the code word
- S represents the syndrome
- I represents the code word
- E represents error.
- the syndrome S is expressed as the product of the error E and the check matrix H, as shown by the following formula (14).
- a clock signal ECK1 is created as NOR of L1 and CKB7 (an inverted signal of CK7) such that ECK1 has H level when L1 is L level.
- Another signal ECK2 is formed in the same manner for L2.
- the number of times at which L1 and L2 take L level corresponds to the numbers of clocks ECK1 and ECK2.
- ⁇ represents an AND circuit represents an OR circuit
- ⁇ o represents a NAND circuit
- o represents a NOR circuit.
- EN1 and EN2 represent the count output of the clock ECK1. With these outputs, it is possible to judge whether the number of the clocks ECK1 is 0, 1 or 2 or greater.
- EN3, EN4 and EN5 are count output of the clocks ECK2. With these signals, it is possible to judge whether the number of clocks ECK2 is 0, 1, 2, 3 or 4 or greater.
- Clocks ECK1 and ECK2 represent not only the number of errors but also the positions of the errors. By setting the flags at the same phases as the clocks ECK1 and ECK2 and determining ANDs of the clocks and the flags, it is possible to count the numbers of errors coinciding with the flags.
- EN6 enables judgment as to whether the number of coincidence between the clocks ECK1 and FLGDD which is the flag outputs set at the same phase as ECK1 is 0 or 1.
- EN7 and EN8 enable judgment as to whether the number of coincidence between ECK2 and FLGDD is 0, 1 or 2.
- EG1 takes H level only in the case of a singular error
- EG2 takes H level only in the case of a double error
- EG3 takes H level in the case of (g), i.e., only when the number of errors exceed the correction capacity.
- FGO to FG5 take H level on conditions of (a) to (f), respectively.
- ERD represent the number of errors
- ERF represent the number of coincidence between the errors and flags as follows.
- An FN output circuit as shown in Fig. 26 is used in order to externally determine whether the flag process is to be conducted with respect to the cases a to q.
- the EN output circuit as shown in Fig. 26 is adapted for counting the number of flags by a counter.
- the output is latched and a comparison is conducted between the latched output and the externally-given value of strategy selection pin NL for each of a to q, whereby a judgment is conducted by ST as to whether the number of the flags received is greater than each value of NL.
- the values of NL corresponding to the conditions to @ to @ are input successively so that the results of comparison between these numbers and the flag number are output by ST in the same sequence and the thus output results are latched by FPCK1 to FPCK7 in accordance with a timing as shown in Fig. 9.
- the latched result of comparison is stored in the next register by EPCK1.
- the outputs from this register are represented by FN1 to FN7.
- the outputs FN1 to FN7 take H level provided that the number of flags received with respect to the respective conditions @ to @ are below the corresponding values of NL and take L level when the flag numbers are equal to or below the values of the corresponding NL.
- NANDs of the error states FGO to FG5 and EG3 and the results FN1 to FN7 of the external strategy selection are obtained and then the NAND of all these NAND outputs is determined as FD (see Fig. 27: FD output circuit).
- a flag output circuit as shown in Fig. 28 is used for the purpose of controlling the flag output by FD.
- the timing of the flag copy and flag raising operations is shown in Fig. 25.
- this embodiment provides a free strategy setting circuit which enables the flag strategy to be set freely, as well as the setting as to whether the flag is to be set up.
- the encoding process and the decoding process have common steps: namely, generation of syndromes and multiplication of the syndromes with constants.
- the constants used in encoding process are encoding constants shown by formulae (28) and (29), while constants used in decoding process are code length correction constants a- n to ⁇ -3n appearing in formula (34).
- the encoding process and the decoding process can be represented by block diagrams as shown in Fig. 29 and 30.
- the pattern creating circuit for use in encoding process may be a circuit which produces EXOR of the syndromes and the encoding constants at every 4 clocks (SO to S3) so that a parity is created for every 4 clocks.
- the pattern creating circuit used in the decoding process also may be a circuit which produces EXOR at every four clocks because it can create an error pattern when two clocks other than ko and Ao 2 /(Ao + A1) are 0 in the output from the processing blocks of formulae (35) to (37). This means that a single pattern creating circuit can be used commonly both for encoding and decoding processes.
- a code word is created by adding the output from the pattern creating circuit to the trailing end of the data I, where the data I is 0 in case of in-3 to in 0, while the pattern creation output is 0 in the cases other than parity.
- correction can be attained by obtaining EXOR with the received word J.
- n represents code length
- H represents check matrix
- I represents code word
- i1 to in-4 represent data
- in-3 to in parity
- n code length
- H check matrix
- J code word
- S syndrome
- I code word
- E error.
- the syndrome S is expressed as the product of the error E and the check matrix H, as shown by the following formula (32).
- PC1 to PC16 are the outputs obtained by shifting PC1 shown in Fig. 34 by means of a shift register.
- constants are assigned to the respective syndromes SI to SIV in accordance with the formula (28) and are output at BUS line Y.
- Fig. 33 shows a circuit which is adapted for conducting EOW control on the thus assigned double-error correction encoding constant outputs and for delivering the result of this control to the input BUS line Y of the multiplier.
- the double-error correction encoding constant output circuit making use of PC1 to PC16 can have a construction as shown in Fig. 35.
- SCL takes L level when the syndrome Sill is created, so that only PC1 to PC8 are used.
- encoding constants corresponding to SI and SII are assigned by means of PC1 to PCB.
- S2 and S3 have no significance, so that 0 is output with respect to S2 and S3.
- the singular-error correction encoding setting circuit can have the same construction and operation principle as the double-error correction encoding circuit, as shown in Figs. 36 and 37. The operation timing of this circuit is shown in Fig. 34.
- the code-length correction constant 1 is created by a circuit having blocks as shown in Fig. 38.
- the operation timing if this circuit is shown in Fig. 39.
- 1 and a- n to a- 3n can be created by an OR circuit.
- the code-length correction circuit may be constructed by a ROM or a later-mentioned exponent/vector conversion circuit, as follows. Namely, the arrangement may be such that ⁇ -n is created by a multiplier in the period of creation of the first syndrome, and then ⁇ - 2n and a- 3n are created by the multiplier, the outputs being latched by a three-state register which in turn is OE-controlled at NCK1 to NCK3.
- An example of such a code-length correction circuit is shown by a block diagram in Fig. 40, while the operation timing of this circuit is shown in Fig. 41.
- encoding and decoding are expressed by D, while the correction capacity is represented by T.
- EOW, EOS and HOE take H level.
- the operation shown in formula (38) can be executed by a circuit shown in Fig. 42.
- KO is extracted from KD by computing AND of KD and KGT and the thus extracted KO is input to a register. Then, the output X- 254 • A0 2 , i.e., A02. (AO + A1) -1 is extracted from the output ZS of the selector shown in Fig. 31, and EXOR of the thus extracted output and the output of the register is computed, whereby an error pattern is created.
- parity is conducted during encoding.
- the above-described operation is conducted only in the period in which syndromes S1 to SW are output from ZS. Other outputs of EP has no meaning. Therefore, by setting PCL at the L level, the output of EP is set at "0". This operation is shown in Fig. 44.
- the operation performed in singular-error correction encoding is shown in Fig. 45.
- the operation of the formula (35) is realized, as in the case of the syndrome creating circuit, multiplying the syndromes Si by a i for k times.
- the operation of formula (36) can be conducted through an EXOR operation.
- the operation of formula (37) can be realized by multiplication of EXOR and A1 2 .
- the formula (37), however, does not matter on the pattern creating operation expressed by the formula (38).
- Operations other than X 2 computation are ordinary operations and, hence, can be realized by a simple circuit.
- the X 2 computation also can be conducted without substantial difficulty by the arrangement shown in Fig. 46.
- the encoding operation is conducted in accordance with the aforementioned formulae (22) to (29). Briefly, the encoding process is to determine syndromes SI to SIV, multiplying these syndromes with encoding constants determined by formulae (28) and (29) and computing EXORs of these outputs for every 4 clocks (SO to S3), whereby parities are generated at every four clocks.
- the encoding operation can be conducted by a circuit having blocks as shown in Fig. 29, including a single multiplication circuit, a syndrome creating circuit capable of outputting syndromes SI to SIV in serial manner, a circuit for serially outputting encoding constants, and an EXOR circuit for computing EXOR for every four clocks.
- the decoding operation is conducted by arithmetic operations in formulae (30) to (39).
- the first step can be conducted by the same construction as the encoding operation, because the operation in this step is to multiply the syndromes S with the code-length correction constants ⁇ -n to a- 3n , through the values of constants are different in both modes.
- the arithmetic operation in accordance with formulae (11) to (20) is conducted by a circuit as shown in Fig. 47. Assuming here that this pattern generating circuit is an EXOR circuit which operates with a period of four clocks, such a circuit can be commonly used both in the encoding and decoding operation.
- a o 2 (A o + A 1 ) -1 can be conducted by a later-mentioned circuit which incorporates a multiplier adapted to conduct multiplication corresponding to four clocks. Since the syndrome creating circuits So to S 3 operate at a period corresponding to four clocks, it is possible to consider four clocks as the basis period. In formulae (30) to (39), the multiplication A 0 •A 2 and the multiplication of y•x 14 used in the division circuit are to be conducted. The code-length correction are conducted in terms of code table and not in the code word basis.
- the operation for multiplying by ⁇ -n to ⁇ -3n is conducted in the period of three clocks so that a time corresponding to one clock is left unused.
- the multiplication Ao-A 2 is conducted in this period.
- the multiplication of y-x' 4 is not conducted only when the code-length correction is to be executed. In such a case, the creation of error pattern is failed with respect to the code word which requires code-length correction. This, however, does not cause any substantial problem if this code word is set in the parity portion.
- the syndrome creating circuit, constant output circuit and the pattern output circuit are the same as those described before, so that detailed description thereof is omitted. A description, therefore, will be started again with respect to the K-creating circuit, comparator circuit, and A-creating circuit.
- the K-creating circuit 481 is a circuit for multiplying the syndrome Si' after code-length correction of formula (34) by ⁇ -i for a time corresponding to the code length, thus executing the operation of formula (35).
- This circuit therefore, relates only to decoding operation and may be realized by a circuit arrangement similar to the syndrome generating circuit. Practically, however, a circuit as shown in Fig. 49 is used because in this case the computation of EXOR with the received word is not necessary.
- a circuit as shown in Fig. 50 is obtained.
- the syndrome SO does not require code-length correction of the formula (34), but syndromes S1 to S3 have to be converted into S1' to S3' through code-length correction.
- the code-length correction is conducted by multiplying the respective syndromes with code-length correction constants.
- Fig. 52 is a block diagram of the comparator circuit.
- the A-creating circuit is composed of a circuit which computes, in accordance with the formula (36), EXOR of KO to K3 created by the K-creating circuit so as to create AO to A2, and a circuit for creating AO + A1, A0 2 and A1 2 .
- the A-creating circuit takes part only in the decoding operation. Because of use of a BUS line type construction, the KO to K3 are not input simultaneously. It is therefore, necessary to latch K by MCK by one clock and computing the EXOR of the output delayed by one clock and K, in order to create A.
- the arrangement of a circuit for realizing such an operation and the timing of operation of this circuit are shown in Figs. 53 and 54, respectively. In Figs.
- x 2 can be determined by a circuit having a construction shown in Fig. 46, as explained before.
- this embodiment enables the scale of the circuit to be reduced by virtue of common use of circuit arrangements both by encoding and decoding and multiplied use of circuit, while maintaining the high speed of encoding and decoding operation and while allowing the correction capacity and the code length to be varied.
- this embodiment contributes to a reduction in the size and improvement in the performance of any device which incorporates this embodiment of the error correction apparatus of the invention.
- Fig. 55 shows the construction of this embodiment.
- a BUS-line structure is used for SO to S3. Since the BUS-line type structure is used for SO to S3, it is necessary that CK and OE (output enable) have to be used in a competing manner in the period corresponding to SO to S3.
- the CK and OE used for the respective one of SO to S3 are controlled by a signal which is shown in Fig. 56.
- Signals CKB1, CKB3, CKB5 and CKB7 are signals obtained through inversion of CK1, CK3, CK5 and CK7, by substituting H with L and vice versa. It is therefore necessary that the input J has to be input for each ji in synchronization with CK1.
- SCL is a signal which takes the L level when the first received word j1 is input. Therefore, SO to S3 are output for every four clocks and the final answer is given in response to each SCL. It is to be understood, however, SPCL is always H in the case of decoding operation.
- an a' circuit is connected between EXOR and the register which outputs J.
- the arrangement also may be such that an ⁇ r-l circuit (l represents any positive number) is placed between EXOR and the register which outputs J, while an at circuit is placed between EXOR and the register of SO and the a circuit.
- Fig. 57 illustrates an arrangement which is suitable for use in the case where multiplication by a' is necessary.
- An a X circuit can be realized by stacking the a circuit shown in Fig. 4 in x-stages.
- the described embodiment offers the advantages pointed out above, while reducing the scale or size of the circuit, even when the number q of the Galois function GF(2q) is large or when the number i of the syndromes Si is large or when the syndromes Si are connected through BUS line.
- the inputs x and y are expressed as follows, with respect to a multiplication circuit on a Galois field GF-(2 8 ).
- the multiplication can be conducted by a circuit having blocks as shown in Fig. 58.
- Fig. 59 shows the construction of the circuit as marked by o in Fig. 58. In the circuit shown in Fig. 58, the value of y is successively multiplied with a by the a circuit shown in Fig.
- Fig. 60 shows a modification intended for minimizing delay time in the gate circuit which is formed by stacking of a circuits in seven stages as explained before. This modification therefore can increase the maximum processing speed.
- y/x is expressed as follows, due to the cyclic nature of the Galois function.
- x 254 can be created by two multiplying operations, i.e., within two clocks. This means that two clocks may be used for two multiplying operations for determining
- X 254 If the requirement is only to determine X 254 , the above-described multiplications are enough. In this case, however, it is necessary that y• X 254 also is to be computed in the period of the fourth clock. To this end, X 14 •y 2 is used in place of X 14 used in the fourth clock. It is assumed that X 14 • y is computed by another multiplication circuit.
- Fig. 61 shows a block diagram of a circuit which is designed for conducting the above-described series of operations.
- This circuit is composed of x 2 , x 4 and x 8 circuits 611, 612 and 613, a multiplier 4 and selections 615, 616, 617 and 618 for selecting the inputs and latches 619, 610 for correcting the delay by the gate.
- the constructions of the x 2 , x 4 and x 8 circuits 611, 612 and 613 are shown in Figs. 63, 64 and 65, respectively.
- a symbol e represents an Exclusive OR circuit, while a symbol represents a BUS line.
- Fig. 62 is a time chart illustrating the operation of the circuit shown in Fig. 61.
- Fig. 66 shows a division circuit having different forms of the x 2 , x 4 and x 8 circuits 61, 62 and 63, which improve the processing speed of the division circuit considerably.
- An example of such a circuit is shown by block diagram in Fig. 67, while the timing of operation of this circuit is shown in Fig. 68.
- outputs ⁇ NO to ⁇ N7•128 are successively obtained at the BUS line Y and are sent to a multiplication circuit 672.
- the output Z of the multiplication circuit 672 successively produces values multiplied with ⁇ NO to ⁇ N7•128 .
- exponent/vector conversion circuit 671 which is adapted to successively output ⁇ NO to ⁇ N7•128 in accordance with binary-coded exponents NO to N7, is shown in Fig. 69.
- this circuit has been constituted by a ROM.
- the described embodiment realizes this circuit by combination of gates.
- a symbol represents an OR circuit.
- the circuit shown in Fig. 67 necessitates 8 clocks for serially outputting NO to N7 till an is formed.
- the arrangement may be modified such that NO to N7 are output in a parallel form and a plurality of multiplication circuits are used to reduce the number of input clocks. In such a modification, however, the size of the circuit will be slightly increased as compared with that of the described embodiment.
Description
- The present invention relates to Galois field arithmetic apparatus, particularly for use in the field of error correction, intended for a reduction of rate of occurrence of error in the path of transmission from a device such as an optical disk device, opto-magnetic disk device, and so forth.
- More particularly, the invention relates to an error correction circuit for use in a digital electronic device such as a digital audio device, particularly a flag strategy setting circuit which is typically a BCH (Bose-Chaudhuri-Hocquenghem) encoding and decoding circuit for use in the error correction circuit.
- Still more particularly, the invention is concerned with a syndrome creating circuit incorporated in the BCH encoding and decoding circuit, particularly to an element-exponent/vector conversion circuit, multiplication circuit and division circuit on a Galois function. The term Galois function in this specification is used to mean a set of a definite number of elements which can be processed by addition, subtraction, multiplication and division and usually expressed as GF(q), where q represents the number of elements.
- Hitherto, error correction apparatus of the type described has been generally designed to operate in response to an error correction code of a predetermined format having factors such as the length of error correction code, correction capacity and interleave. Thus, the known error correction apparatus could not operate when the correction code of a different format is applied and when a plurality of blocks of correction codes of various formats are supplied serially.
- Optical disks and opto-magnetic disks are usually kept under sector administration. The sector contains not only the ordinary data but also address data. Protection of these two types of data is a matter of great significance. Conventionally, error correction codes such as read solomon codes have been used as the protection means for protecting the ordinary data portion, while error detection code such as CRC (Cyclic Redundancy Check) codes have been used as means for protecting the address data. An erroneous recognition of the address data will lead to an error of the whole sector. Therefore, in the conventional systems, a greater importance has been given to the detection accuracy than to the correction performance insofar as the address data is concerned. Thus, any error in the address data has been corrected through an over-writing or re-try. There are some types of recording medium which enables the address data to be corrected through re-try but not through overwriting. An example of such types of recording medium is opto-magnetic disk. In the correction of address data in such a recording medium, it is necessary that the disk is rotated three times: namely, first rotation for erasure, second rotation for writing and third rotation for check, each rotation requiring detection of address. In consequence, the number of the re-try cycles is increased resulting in a longer delay time, particularly in the medium having a greater rate of occurrence of error. It would be possible to use an error correction code for the purpose of correction of address data. In general, however, error correction device requires a much more complicated construction than the error detecting device. When the correction of address data is to be conducted by means of correction code, it is necessary to use a pair of error correction device: namely, one for correction. of address data and one for correction of ordinary data.
- THE THEORY OF ERROR CORRECTING CODES, written by F. J. Macwiliam and N. J. A. Sloan, published in 1978 by North Holland Publishing Company, U.S.A., proposes CIRC (Cross-Interleaved Read Solomon Code), particularly two-staged decoding having 2-symbol error correction facility. Typical examples of the strategy used in this two-staged decoding are as follows, where C1 decoder and C2 decoder mean the decoders which conduct decoding earlier and later, respectively.
-
-
- In the case of a digital audio equipment, any detection error causes a click noise and, hence, is critical. Any error which cannot be corrected, however, can be compensated for by interpolation to such an extent that it does not substantially impair the audio effect, provided that the error is detected. It is therefore necessary to maximize the error detection capacity at the cost of reduced error correction capacity, as in the simple strategy (1) shown above. When the compensation by, for example, the interpolation is not effective, 2 symbol error correction is performed by the C1 decoder as in the case of the super strategy (2) mentioned above. In this case, flag is set up to enhance also the error detection capacity, thereby to reduce the rate of error detection failure.
- It is thus possible to optimize the format for error detection by selectively changing the strategy, even when the format has an identical construction both in vertical and horizontal directions.
- A description will be made hereinunder as to a super strategy (2) on an assumption that C1 decoder (32, 28) and C2 decoder (28, 24) are used.
- The following process is conducted in accordance with the result of judgment of received syndrome Sc1.
- (a) Sc1 = "0" → Correction not conducted
- ; Fc1 = 0
- (b) Sc1 = "1 " → 1 symbol error corrected
- ; Fc1 = 0
- (c) Sc1 = "2"→ 2 symbol error corrected
- ; Fc1 = 1
- (d) Sc1 ≥ "3"→ Correction not conducted
- ; Fc1 = 1
where, - "n": n-symbol error syndrome
- Fc1: Flag data created in C1 decoder and sent to C2 decoder
- Fc1 = 0 → No error
- Fc1 = 1 - Possibility of inclusion of any error
- ii) C2 decoder
- ; Fc1 = 1
- The following process is conducted in accordance with the result of judgment of received syndrome Sc2, and number and positions of flags Fc1 from C1 decoder.
- (a) Sc2 = "0" → Correction not conducted
- ; Fc2 = 0
- (b) Sc2 = "1 " → 1 symbol error corrected
- ; Fc2 = 0
- (c) Sc1 = "2",
Nc1 4, Lc1 = 2- - 2 symbol error corrected
- ; Fc2 = 0,
-
Nc1 3, Lc1 = 1 orNc1 2, Lc1 = 0 - - correction not conducted
- ; Fc2 = 1,
- Other cases
- - Correction not conducted
- ; Fc2 = Fc1
- (d) Sc2 ≥ "3", Nc1 ≤ 2
- - Correction not conducted
- Fc2 = 1
- Other cases
- → Correction not conducted
- Fc2 = Fc1
where,- Nc1: Number of flags Fc1 received by C2 decoder
- Lc1: Number of flags Fc1 coincided with error location
- Fc2: Flag created in C2 decoder Fc2 = 0 → No error Fc2 = 1 - All errors Fc2 = Fcl - Flags created in C1 decoder are copied.
- The above-described principle will be more readily understood from Fig. 2. It will be seen that the simple strategy is the strategy in which the C2 decoder has the same construction as the C1 decoder shown in Fig. 2(1
- Conventionally, once the flag strategy in each of the C1 and C2 decoders is set as described above, the thus set strategy is fixed and cannot be freely changed or adjusted.
- It is also a conventional way to conduct the BCH encoding by means of a division circuit employing a creating polynominal and to conduct the decoding by means of circuits of the respective algorithms such as those according to Peterson's method or Bharenkamp-Massey.
- Thus, the conventional system employs different circuits for encoding and decoding. Attempts have been made to conduct encoding and decoding by making use of the same circuit in a multiplied manner in accordance with, for example, microprogramming. In such a case, however, it is necessary that delicate difference be made between the encoding process and the decoding process. Therefore, when the encoding and decoding are to be conducted by means of the same circuit board or chip, the number of circuits or the capacity of ROM is increased inevitably. This in turn causes the size of the device such as an optical disk device to be increased undesirably.
-
- Namely, in case of double error correction encoding, the syndrome S is created as the product of a check matrix H and the received word J, as shown by the formula (2).
-
-
- Thus, the syndrome creating circuit on the Galois function GF(2q) can be constructed as shown in Fig. 3. The ai circuit can be realized by stacking the a circuit of Fig. 4 in i stages, on condition of a primitive polynominal p(x) = X8 + X4 + X3 + X2 + 1. Thus, the syndrome creating circuit for realizing SO to S3 expressed by the formula (2) can be constructed as shown in Fig. 5. In Fig. 5, CK represents clock for each received word Ji, while CL represents the clear for each code length.
- Obviously, the size of the construction shown in Fig. 5 becomes large as the number q or the number i becomes large. In addition, the arrangement shown in Fig. 5 is not suitable for use in the case where the syndromes SO to S3 are interconnected and processed through a BUS line.
- There are two types of ways of expression of the element of the Galois field: namely, expression in terms of vector and expression in terms of exponent. Representing a Galois field having q elements by GF-(q), the element a8 created on GF(28) from the a, which is the root of the primitive polynominal p(x) = x8 + x 4 + X3 + x 2 + 1 is expressed as follows.
-
- The VE (vector to exponent) conversion and EV (exponent to vector) conversion can be conducted by making use of a ROM.
- When a division is to be completed in the period of 1 (one) clock as shown in Fig. 6, it is necessary to employ 3 (three) ROMs. In contrast, when the division is conducted by employing one ROM for VE conversion and one ROM for EV conversion as shown in Fig. 7, the operation requires a period corresponding to two clocks: first clock period for latching by means of a register and second clock period for adding a.
- Similarly, when a multiplication is to be completed in the period of 1 (one) clock as shown in Fig. 8, it is necessary to employ 3 (three) ROMs, whereas, when the multiplication is conducted by employing one ROM for VE conversion and one ROM for EV conversion as shown in Fig. 9, the operation requires a period corresponding to two clocks: first clock period for latching a by means of a register and second clock period for adding b.
- It would be possible to conduct division or multiplication of two elements expressed in term of vectors directly by a ROM constituting a VE or EV conversion circuit. In such a case, the ROM will be required to have an impractically large capacity, particularly when the number of elements is large. A prior art multiplication circuit is shown in EP-0129849-A3 which employs ROMs for VE and EV conversion.
- It is an object of the present invention to provide an error correction circuit having an arithmetic circuit which is small in size but yet capable of conducting division and multiplication, as well as VE and EV conversion, of unknowns of Galois function, without using any ROM.
- EP-A-0080528 discloses a Galois field multiplication apparatus implemented with a logic circuit rather than ROM.
- Accordingly, the present invention provides multiplication apparatus (Fig 55) for multiplying two arbitrary elements X and Y on a Galois Field GF(2m), comprising;
- means for calculating γ-ai for each integer i from 0 to m - 1 inclusive, where a is a primitive element on the Galois Field GF(2m);
- means for multiplying each component x of X with the corresponding term γ-ai thus calculated; and
- means for generating the binary sum of all the products thus multiplied.
- In another aspect the invention provides a method of digital signal processing which includes the step of multiplying together two arbitrary elements X, Y on a Galois Field GF(2m) which comprises the steps of;
- calculating y•-αi for each integer i from 0 to m - 1 inclusive, where a is a primitive element y on the field GF(2m);
- multiplying each y•αi thus calculated with a corresponding component xi; and
- generating the binary sum of all products thus multiplied.
- Other prefered features and embodiments of the invention will be apparant from the following description and claims. The invention will now be described, by way of example only, with reference to the accompanying drawings, in which;
-
- Fig. 1 is an illustration of a encoding/decoding circuit having erasure correction and error correction function, in a first embodiment of the error correction apparatus of the present invention;
- Fig. 2 is an illustration of super strategy;
- Fig. 3 is an illustration of a syndrome creating circuit for creating syndromes Si;
- Fig. 4 is an illustration of a a circuit;
- Fig. 5 is an illustration of a known syndrome creating circuit;
- Figs. 6 and 7 are illustrations of construction of a known division circuit;
- Figs. 8 and 9 are illustrations of a known multiplication circuit;
- Fig. 10 is an illustration of an interleave;
- Fig. 11 is a time chart illustrating the operation starting with data-in and ending in data-out after error correction;
- Fig. 14 is an illustration of a decoding hardware;
- Fig. 15 is an illustration of a encoding hardware;
- Fig. 16 is a flow chart showing the flow of computation;
- Fig. 17 is an illustration of the construction of a encoding/decoding unit in the first embodiment of the invention;
- Fig. 18 is an illustration of erasure correction and error correction hardware in the first embodiment of the invention;
- Fig. 19 is an illustration of operation in the case of an interleave;
- Fig. 20 is an illustration of the position of error correction;
- Fig. 21 is an illustration of free strategy in the first embodiment of the invention;
- Fig. 22 is a table showing the state of error used in double-error correction;
- Fig. 23 is a table showing the state of error used in singular-error correction;
- Fig. 24 is an illustration of the construction of an EN output circuit in a second embodiment of the present invention;
- Fig. 25 is an operation timing chart;
- Fig. 26 is an illustration of the construction of an FN output circuit in a second embodiment of the present invention;
- Fig. 27 is an illustration of the construction of an DN output circuit in a second embodiment of the present invention;
- Fig. 28 is an illustration of the construction of a flag output circuit in accordance with GCL control;
- Fig. 29 is a block diagram of a encoding circuit;
- Fig. 30 is a block diagram of a decoding circuit;
- Fig. 31 is a block diagram of a encoding/decoding circuit in a third embodiment of the present invention;
- Fig. 32 is a timing chart of operation of a syndrome creating circuit in the encoding mode;
- Fig. 33 is a block diagram of a double-error correction encoding constant circuit;
- Fig. 34 is a timing chart showing the operation of the encoding constant circuit;
- Fig. 35 is an illustration of a double-error correction encoding constant output circuit;
- Fig. 36 is a block diagram of a singular-error correction encoding constant circuit;
- Fig. 37 is an illustration of a singular-error correction encoding constant output circuit;
- Fig. 38 is a code-length correction constant circuit;
- Fig. 39 is a timing chart showing the operation of the code-length correction constant circuit;
- Fig. 40 is a block diagram of a variable-code-length correction circuit;
- Fig. 41 is a timing chart showing the operation of the variable-code-length correction circuit;
- Fig. 42 is a block diagram of a pattern creating circuit;
- Fig. 43 is a timing chart showing the operation of a pattern creating circuit in decoding mode;
- Figs. 44 and 45 are timing charts showing the operation of the pattern creating circuit in encoding mode;
- Fig. 46 is an illustration of an X2 circuit;
- Fig. 47 is a block diagram of a encoding circuit;
- Fig. 48 is a block diagram of encoding/decoding circuit in a modification of the third embodiment;
- Fig. 49 is an illustration of a K-creating circuit;
- Fig. 50 is an illustration of a K-creating circuit for BUS line control;
- Fig. 51 is an illustration of timing of operation of the K-creating circuit;
- Fig. 52 is a block diagram of a comparator circuit;
- Fig. 53 is a block diagram of an A-creating circuit;
- Fig. 54 is an illustration of timing of operation of the A-creating circuit;
- Fig. 55 is an illustration of a syndrome creating circuit in a fourth embodiment of the present invention;
- Fig. 56 is an illustration of a timing signal;
- Fig. 57 is an illustration of another example of the syndrome circuit used in the error correction apparatus of the present invention;
- Fig. 58 is an illustration of a multiplication circuit used in a fifth embodiment of the present invention;
- Fig. 59 is an illustration of the construction of an AND circuit incorporated in the circuit shown in Fig. 58;
- Fig. 60 is an illustration of an improvement in the multiplication circuit;
- Fig. 61 is a block diagram of a division circuit used in the fifth embodiment of the invention;
- Fig. 62 is a time chart showing the operation of the fifth embodiment;
- Fig. 63 is an illustration of the construction of an X2 circuit in the division circuit;
- Fig. 64 is an illustration of the construction of an X4 circuit;
- Fig. 65 is an illustration of the construction of an X8 circuit;
- Fig. 66 is an illustration of an improvement in the division circuit;
- Fig. 67 is a block diagram of an exponent/vector conversion circuit used in the fifth embodiment of the invention;
- Fig. 68 is a timing chart showing the operation of the circuit shown in Fig. 67; and
- Fig. 69 is an illustration of the construction of an exponent/vector circuit.
- A first embodiment of the present invention will be described hereinunder with reference to the accompanying drawings. A description will be made. first as to decoding of an error correction code. A read solomon code (referred to as "RS code" hereinafter) as an example of error correction signal typically has a cross-interleaved construction as shown in Fig. 10. This means that each block requires two times of error correction as represented by C1 and C2. When input data is supplied without any interval, the processing delay time will be accumulated unless an erasure correction and error correction unit is provided for each of C1 and C2. order to simplify the construction of the hardware, therefore, this embodiment is constructed such that a single erasure correction and error correction unit is made to operate at a doubled speed as denoted by 10 and 02 in Fig. 11 when the input data is supplied without interval. The erasure correction and error correction unit, however, operates at a normal speed in the when there is no interleave or when the data transmission speed is high. Referring to Fig. 11, data is input as represented by data-in in response to gate-in and, after a computation for error correction is completed, the corrected data is output as represented by data-out. As will be understood from this Figure, the data input/output time in the doubled
speed mode - Therefore, the erasure correction and error. correction unit has, as shown in Fig. 12, a
RAM 121 for storing the input data, an erasure correct and errorcorrect block 122 and acontrol block 123 for varying the timing signal for driving theblock 122 in accordance with an input format select signal. - Amongst various error correction methods proposed hitherto, this embodiment makes use of an algorithm as shown in Fig. 13, which can simplify the hardware construction and, hence, facilitates the operation under the condition where the timing signal is variable. Fig. 14 shows the construction of decoding hardware adopted when this algorithm is used.
- Usually, the encoding is conducted by dividing data by a creating polynominant. In this embodiment, however, the encoding is conducted by varying the control of the error correction unit which is used in decoding. The construction of a hardware used for this purpose is shown in Fig. 15.
- Fig. 16 shows the flow of the process for creating parity bit.
- The decoding and encoding described above can be united with each other by a hardware as shown in Fig. 17.
-
-
-
-
- The factor A-1 can be definitely determined provided that i, j, k and ℓ are given.
-
- Fig. 18 is a block diagram of the erasure correction and error correction circuit. Since i, j, k and I are definite, the A-1 creating circuit can be determined by several ROMs. Therefore, the encoding/decoding circuit including the erasure correction function can be constructed as shown in Fig. 1.
- In order to enable the operation to be switched between ordinary encoding/decoding mode and the erasure correction mode, it is necessary to conduct a switching between a
ROM 11 and aROM 12 in accordance with an erasure correction select signal ER. To this end, state-controlling ROMs are used as theROMs - The correcting operation is delayed behind the generation of syndrome by a period corresponding to one received word. The received word, therefore, has to be temporarily stored in
buffer 13. The error pattern computed at the time of generation of syndrome has to be temporarily stored in the addresses i, j, k and I of a buffer 54, in order that the pattern is output at the positions i, j, k and I in synchronization with the output from thebuffer 13. When i, j, k and I is output from the AD, this value is output so that the error pattern is output in synchronization with the output from thebuffer 13. - When there is no interleave in the error correction code or when the processing is conducted at the doubled speed, this device is used only for a single error correction. In such a case, this device can be used as it is without any modification.
- In the case where the error correction code is interleaved, this device is used for two times of error correcting operation. When the interleaved construction of error correction code is adopted, error correction is conducted both in the vertical and horizontal directions as shown by arrows in Fig. 10. The operation of the device at the doubled speed can be conducted by doubling the rates of clock and trigger. In this case, however, a difficulty is encountered due to difference in the format between C1 and C2. The format, however, can be varied by imparting different values of the correction capacity T and code lengths n, k to C1 and C2.
- Fig. 19 shows a circuit which is adopted when the error correction code has an interleaved structure. When n1, n2, T1 and T2 have been set, it is possible to obtain different values of n. T and k for C1 and C2, by changing the level of the select signal C. For instance, different error correction capacities and code lengths are obtained for C1 and C2, by setting the level of the selection signal C as C = 0 for C1 and C = 1 for C2, respectively. The code length k can be determined by n-2. T-E. It is thus possible to conduct the desired interleaved processing.
- It is also possible to select processing in vertical and horizontal directions of the format by selectively setting the signal C as D - 1 or 1 - D.
- As will be understood from the foregoing description, according to this embodiment of the invention, an error correction apparatus can be driven for general purposes. By constructing this apparatus in one chip and designing the same to be adaptable to a variety of formats, it becomes possible to apply the invention not only to optical disk and opto-magnetic disk devices but also to all types of transmission path which suffers from large rate of occurrence of error.
- It is also possible to prepare a plurality of such chips and to connect them in an interleaved manner, so as to construct an apparatus which can be applied to systems in which the rate of occurrence of error and the processing speed are specifically high.
- Referring to Fig. 20, a mark S.M. represents a sector mark which indicates the beginning end of a sector. ID is read in accordance with clock component extracted by SYNC. A symbol A.M. in ID represents an address mark which indicates the beginning of an address. GAP is formed between ID section and DATA section for the purpose of absorbing jitter, thereby to accommodate any offset of ID section from data section including SYNC, S.M. and DATA. The same applies also to the data section. The last GAP is preserved for the purpose of absorbing any offset between the present sector and the next sector. This data structure can be processed by the apparatus of the present invention, even though the address mark and the data section are coded in different formats.
- As will be understood from the foregoing description, according to the invention, an error correction apparatus having a general-purpose format selector section and an erasure correction selecting section including erasure position input section is constructed in a single chip so that a single erasure correction and error correction apparatus can operate for a variety of formats.
- In addition, when the input data has a plurality of different formats, error correction can be conducted by a single erasure correction and error correction block by employing a select signal for selecting different formats.
- It is also to be noted that, in the described embodiment, a control unit may be employed which enables a single erasure correction and error correction block to operate at a doubled speed when the data to be corrected has an interleaved structure and at a normal speed when the error correction code does not have any interleaved structure.
- Furthermore, the apparatus of the described embodiment, when applied to erasure correction and error correction of data having a sector construction, the error correction can be conducted not only on the data in the sector but also on address data of the sector by means of the same erasure correction and error correction block.
- A second embodiment of the present invention will be described hereinunder. In the encoding and decoding of BCH code, the fixed
strategy - ⓐ Sci = "0"
- ⓑ Sci = "1", Lci = 1
- © Sci = "1 ", Lci = 0
- ⓓ Sci = "2", Lci = 2
- ⓔ Sci = "2", Lci = 1
- ⓕ Sci = "2", Lci = 0
- ⓖ Sci = "3"
where,- Sci: syndrome received by Ci decoder
- Lci: Number of preceding flags Fci coincided with the error locations of Ci decoder
- Nci: Number of flags Fci received by Ci decoder
- Flags from the preceding device such as MODEM can be utilized provided that both the C1 and C2 decoders are of free strategy type. Whether the flag is to be copied or set up is controlled by another signal GCL. To this end, when the condition of GCL = H is met, the value of the flags created by the Ci decoder is determined as follows.
- Fcx = 0 : No error
- Fcx = Fci : Flags created in the preceding decoder are copied.
- In order to turn the Fci and "0" into "1 " as the value of Fcx, it is necessary that the condition of GCL = L is met under the required condition.
- This principle will be more readily understood from the following description taken in conjunction with Fig. 21.
- In order to realize the function explained above, a circuit as shown in Fig. 24 is constructed by using L1 and L2 of an algorithm which follows formulae (12) to (21) shown below. The timing of operation of this circuit arrangement is shown in Fig. 25.
-
-
- It is assumed here that errors ei and ej exist at positions i and j, respectively.
-
-
-
-
-
-
- ① When there is no error (ei = ej = 0)
- L1 = 0
- L2 = 0
- e = 0
- ② In case of singular error (
ei # 0, ej = 0) -
- ③ In case of double error (
ei # 0, ej # 0)- L1: indefinite
- L2: L2 = 0 only on conditions of k = i and k = j
- e: e = ei on condition of k = i and e = ej on condition of k = j
- A clock signal ECK1 is created as NOR of L1 and CKB7 (an inverted signal of CK7) such that ECK1 has H level when L1 is L level. Another signal ECK2 is formed in the same manner for L2. The number of times at which L1 and L2 take L level corresponds to the numbers of clocks ECK1 and ECK2. The numbers of clocks ECK1 and ECK2 given by the formula (21) is shown in Figs. 22 and 23. More specifically, Fig. 22 shows the clock numbers as obtained when the correction capacity T is set as T = 2, while Fig. 23 shows the clock numbers as obtained under the condition of T = 1. When T is set as T = 1, the signal created by L2 is meaningless so that it is crossed out by oblique line in Fig. 23. The numbers of clocks ECK1 and ECK2 would be conducted without difficulty by using a counter and a comparator. In this embodiment, however, an EN output having a construction as shown in Fig. 24 is used in the judgment of conditions (a) to (g) in order to reduce the scale of the circuit. In this Figure, a mark ⊚ represents an AND circuit represents an OR circuit, ⊚o represents a NAND circuit, and o represents a NOR circuit.
- EN1 and EN2 represent the count output of the clock ECK1. With these outputs, it is possible to judge whether the number of the clocks ECK1 is 0, 1 or 2 or greater. EN3, EN4 and EN5 are count output of the clocks ECK2. With these signals, it is possible to judge whether the number of clocks ECK2 is 0, 1, 2, 3 or 4 or greater. Clocks ECK1 and ECK2 represent not only the number of errors but also the positions of the errors. By setting the flags at the same phases as the clocks ECK1 and ECK2 and determining ANDs of the clocks and the flags, it is possible to count the numbers of errors coinciding with the flags. EN6 enables judgment as to whether the number of coincidence between the clocks ECK1 and FLGDD which is the flag outputs set at the same phase as ECK1 is 0 or 1. EN7 and EN8 enable judgment as to whether the number of coincidence between ECK2 and FLGDD is 0, 1 or 2.
- Before arrival of the clocks ECK1 and ECK2 created by the next received word, it is necessary to clear the output by ECL1 so as to re-start the counting. To this end, it is necessary that the output before clearing by ECL1 is stored in the register of the next stage at a timing of a signal EPCK1 shown in Fig. 25. Thus, the processing of flags and error correction are conducted in accordance with the output from the register of the next stage. The gate outputs EG1 to EG3 and FGO to FG5 representing the states of errors shown in Figs. 22 and 23 are expressed as follows, by using EN1 to EN8.
- EG1 = (T1 + T2)•(EN5 + T2). EN1• ENB2
- EG2 = T2•EN4• ENB5
- EG3 = T1• ENB1 • ENB2 + T2• ENB4. ENB5
- FGO = T1•EN2 + T2• EN2. EN5
- FG1 = EN6•EG1
- FG2 = ENB6-EG1
- FG3 = ENS• EG2
- FG4 = EN7. EG2
- FG5 = ENB7• ENB8• EG2
- Thus, EG1 takes H level only in the case of a singular error, while EG2 takes H level only in the case of a double error. Note that EG2 takes L level whenever the correction capacity T is set as T = 1. EG3 takes H level in the case of (g), i.e., only when the number of errors exceed the correction capacity. FGO to FG5 take H level on conditions of (a) to (f), respectively. In Figs. 22 and 23, ERD represent the number of errors, while ERF represent the number of coincidence between the errors and flags as follows.
- ERD1 = EG1 + EG3
- ERD2 = EG2 + EG3
- ERF1 = FG1 + FG4
- ERF2 = FG3
- An FN output circuit as shown in Fig. 26 is used in order to externally determine whether the flag process is to be conducted with respect to the cases ⓐ to ⓠ.
- The EN output circuit as shown in Fig. 26 is adapted for counting the number of flags by a counter. The output is latched and a comparison is conducted between the latched output and the externally-given value of strategy selection pin NL for each of ⓐ to ⓠ, whereby a judgment is conducted by ST as to whether the number of the flags received is greater than each value of NL. The values of NL corresponding to the conditions to @ to @ are input successively so that the results of comparison between these numbers and the flag number are output by ST in the same sequence and the thus output results are latched by FPCK1 to FPCK7 in accordance with a timing as shown in Fig. 9.
- The latched result of comparison is stored in the next register by EPCK1. The outputs from this register are represented by FN1 to FN7. Thus, the outputs FN1 to FN7 take H level provided that the number of flags received with respect to the respective conditions @ to @ are below the corresponding values of NL and take L level when the flag numbers are equal to or below the values of the corresponding NL.
- Then, NANDs of the error states FGO to FG5 and EG3 and the results FN1 to FN7 of the external strategy selection are obtained and then the NAND of all these NAND outputs is determined as FD (see Fig. 27: FD output circuit).
- A flag output circuit as shown in Fig. 28 is used for the purpose of controlling the flag output by FD. The flag output circuit produces AND of FD and the flag FLGI and latches the AND output as FLGO. It will be seen that the AND output is delivered as it is when GCL is externally set at H level but is delivered as FLGO = H when GCL is externally set at L level. As a result, flag copy is executed on condition of GCL = H and while an operation for setting the flag up is executed on condition of GCL = L. The timing of the flag copy and flag raising operations is shown in Fig. 25.
- As will be understood from the foregoing description, this embodiment provides a free strategy setting circuit which enables the flag strategy to be set freely, as well as the setting as to whether the flag is to be set up.
- A description will be made hereinunder as to a third embodiment of the present invention, particularly to encoding and decoding circuit of an error correction apparatus suitable for use in an optical disk device DAT or a similar device.
- It is assumed here that the encoding process is conducted in accordance with the following formulae (22) to (29), while decoding is conducted in accordance with the following formulae (30) to (39). The encoding process and the decoding process have common steps: namely, generation of syndromes and multiplication of the syndromes with constants. The constants used in encoding process are encoding constants shown by formulae (28) and (29), while constants used in decoding process are code length correction constants a-n to α-3n appearing in formula (34). The encoding process and the decoding process can be represented by block diagrams as shown in Fig. 29 and 30. The pattern creating circuit for use in encoding process may be a circuit which produces EXOR of the syndromes and the encoding constants at every 4 clocks (SO to S3) so that a parity is created for every 4 clocks. The pattern creating circuit used in the decoding process also may be a circuit which produces EXOR at every four clocks because it can create an error pattern when two clocks other than ko and Ao2/(Ao + A1) are 0 in the output from the processing blocks of formulae (35) to (37). This means that a single pattern creating circuit can be used commonly both for encoding and decoding processes. In the coding process, a code word is created by adding the output from the pattern creating circuit to the trailing end of the data I, where the data I is 0 in case of in-3 to in 0, while the pattern creation output is 0 in the cases other than parity. In the decoding process, correction can be attained by obtaining EXOR with the received word J.
-
- where, n represents code length, H represents check matrix, I represents code word, i1 to in-4 represent data and in-3 to in represent parity.
-
-
-
-
-
-
-
- wherein, n represents code length, H represents check matrix, J represents code word, S represents syndrome, I represents code word and E represents error.
-
- It is assumed here that errors ei and ej exist at positions i and j, respectively.
-
-
-
-
-
-
-
- ① When there is no error (ei = ej = 0)
- L1 = 0
- L2 = 0
- e = 0
- ② In case of singular error
- L1: L1 = 0 only when k = i
- L2 = 0
- e: e = ei only when k = i
- ③ In case of double error (
ei # 0, ej # 0)- L1: indefinite
- L2: L2 = 0 only on conditions of k = i and k = j
- e: e = ei on condition of k = i and e = ej on condition of k = j
- It is therefore possible to selectively conduct encoding and decoding processes by designing an encoding and decoding circuit in which, as shown in Fig. 31, an encoding circuit is used as an unknown and a constant output circuit is formed by adding selector and encoding
constant output circuit 291 to a code length correctionconstant generating circuit 301. Each of the blocks can have a simplified construction as follows. - When a single multiplier is used as in Fig. 31, it is necessary that the signal output from the syndrome generating circuit is conducted in a serial manner by means of a BUS line. In case of encoding, the serial output can be conducted provided that the input in-3 to in corresponding to the parity portion is 0. However, if the input in-3 to in is not zero, the following operation is necessary.
- In order to realize S1 to SIV shown in formula (28) in double-error correction encoding, the syndrome creation circuit should be operated with the input data in-3 to in set at 0. To this end, the clear input to be supplied to the register which latches the received word J is held at L level during the period corresponding to in-3 to in. Therefore, syndromes SI [SO, S1, S2, S3] are formed by the words received in the period between i1 and in-4, when in-3 is being input. The syndrome creating circuit continues to operate even during receiving in-2 = 0 so that syndromes SII [SO, α•S1, α2•S2, α3•S3] are created. Similarly, when in - 1 = 0 and in = 0 are input, syndromes Sill [SO, a2. S1, a4. S2, a6. S3] and SIV [SO, a3. S1, α6• S2, α9• S3] are formed.
- The same applies also to the case of singular error. Namely, when in - 1 = 0 is being input, syndromes SI [SO, S1] are created in accordance with inputs in the period between i1 to in-2, and, when the next input in = 0 is being received, syndromes SII [SO, α•S1] are created. Thus, the encoding syndrome creation circuit can be realized simply by controlling the SPCL of the decoding syndrome generating circuit. The timing of this control is shown in Fig. 32.
- In the double-error correction encoding, after determination of the syndromes SI to SIV by the syndrome creating circuit, it is necessary to multiply the syndromes with double-error correction encoding constant, in order to create the parity in-3 to in. As shown in formulae (28) and (29), a single encoding constant is used fixedly in each of the singular error correction and double-error correction. Thus, the multiplication can be conducted by a circuit which multiplies the syndromes with the corresponding constant in synchronization with these syndromes. Fig. 33 shows a block diagram of an example of such a circuit.
- PC1 to PC16 are the outputs obtained by shifting PC1 shown in Fig. 34 by means of a shift register. By the outputs PC1 to PC16, constants are assigned to the respective syndromes SI to SIV in accordance with the formula (28) and are output at BUS line Y. Fig. 33 shows a circuit which is adapted for conducting EOW control on the thus assigned double-error correction encoding constant outputs and for delivering the result of this control to the input BUS line Y of the multiplier. The double-error correction encoding constant output circuit making use of PC1 to PC16 can have a construction as shown in Fig. 35.
- In case of singular-error correction encoding operation, SCL takes L level when the syndrome Sill is created, so that only PC1 to PC8 are used. Namely, encoding constants corresponding to SI and SII are assigned by means of PC1 to PCB. In this case, S2 and S3 have no significance, so that 0 is output with respect to S2 and S3. It will be understood that the singular-error correction encoding setting circuit can have the same construction and operation principle as the double-error correction encoding circuit, as shown in Figs. 36 and 37. The operation timing of this circuit is shown in Fig. 34.
- As in the case of the encoding constant, the code-length correction constant 1 is created by a circuit having blocks as shown in Fig. 38. The operation timing if this circuit is shown in Fig. 39. In the case where the length is fixed, 1 and a-n to a-3n can be created by an OR circuit.
- When the length n is variable, the code-length correction circuit may be constructed by a ROM or a later-mentioned exponent/vector conversion circuit, as follows. Namely, the arrangement may be such that α-n is created by a multiplier in the period of creation of the first syndrome, and then α- 2n and a-3n are created by the multiplier, the outputs being latched by a three-state register which in turn is OE-controlled at NCK1 to NCK3. An example of such a code-length correction circuit is shown by a block diagram in Fig. 40, while the operation timing of this circuit is shown in Fig. 41.
- As to the selection signal, encoding and decoding are expressed by D, while the correction capacity is represented by T. EOW (encoding, T = 2), EOS (encoding, T = 1) and HOE (decoding) are executed by the setting of T and D. When T and D are not set, EOW, EOS and HOE take H level.
- The operation shown in formula (38) can be executed by a circuit shown in Fig. 42. KO is extracted from KD by computing AND of KD and KGT and the thus extracted KO is input to a register. Then, the output X-254 • A02, i.e., A02. (AO + A1)-1 is extracted from the output ZS of the selector shown in Fig. 31, and EXOR of the thus extracted output and the output of the register is computed, whereby an error pattern is created. In other cases, EXOR is computed by setting the output as 0, by applying a condition of KGT = XGT = 0. This operation is commonly adopted in singular and double error correction. The timing of operation of this circuit is shown in Fig. 43. In case of decoding, however, a condition of PCL = H applies.
- The creation of parity is conducted during encoding. In the case of double-error correction coding, values obtained by multiplying the syndromes SI to SIV by encoding constants are derived from ZS. It is therefore possible to generate parities in-3 to in successively, by continuously producing EXOR of the ZS output and the output of the register cleared by CKB6 on condition of XGT = H and KGT = L until the register output is cleared again by CKB6. The above-described operation is conducted only in the period in which syndromes S1 to SW are output from ZS. Other outputs of EP has no meaning. Therefore, by setting PCL at the L level, the output of EP is set at "0". This operation is shown in Fig. 44. The operation performed in singular-error correction encoding is shown in Fig. 45.
- The operation of the formula (35) is realized, as in the case of the syndrome creating circuit, multiplying the syndromes Si by ai for k times. The operation of formula (36) can be conducted through an EXOR operation. The operation of formula (37) can be realized by multiplication of EXOR and A12. The formula (37), however, does not matter on the pattern creating operation expressed by the formula (38). Operations other than X2 computation are ordinary operations and, hence, can be realized by a simple circuit. The X2 computation also can be conducted without substantial difficulty by the arrangement shown in Fig. 46.
- A description will be made hereinunder as to a modification of this embodiment.
- The encoding operation is conducted in accordance with the aforementioned formulae (22) to (29). Briefly, the encoding process is to determine syndromes SI to SIV, multiplying these syndromes with encoding constants determined by formulae (28) and (29) and computing EXORs of these outputs for every 4 clocks (SO to S3), whereby parities are generated at every four clocks. Thus, the encoding operation can be conducted by a circuit having blocks as shown in Fig. 29, including a single multiplication circuit, a syndrome creating circuit capable of outputting syndromes SI to SIV in serial manner, a circuit for serially outputting encoding constants, and an EXOR circuit for computing EXOR for every four clocks.
- The decoding operation is conducted by arithmetic operations in formulae (30) to (39). The first step can be conducted by the same construction as the encoding operation, because the operation in this step is to multiply the syndromes S with the code-length correction constants α-n to a-3n, through the values of constants are different in both modes. Then, the arithmetic operation in accordance with formulae (11) to (20) is conducted by a circuit as shown in Fig. 47. Assuming here that this pattern generating circuit is an EXOR circuit which operates with a period of four clocks, such a circuit can be commonly used both in the encoding and decoding operation.
- A discussion will be made hereinunder as to an attempt for combining three multipliers in Fig. 47 into a single multiplier.
- The division of Ao 2(Ao + A1 )-1 can be conducted by a later-mentioned circuit which incorporates a multiplier adapted to conduct multiplication corresponding to four clocks. Since the syndrome creating circuits So to S3 operate at a period corresponding to four clocks, it is possible to consider four clocks as the basis period. In formulae (30) to (39), the multiplication A0•A2 and the multiplication of y•x14 used in the division circuit are to be conducted. The code-length correction are conducted in terms of code table and not in the code word basis.
- The operation for multiplying by α-n to α-3n is conducted in the period of three clocks so that a time corresponding to one clock is left unused. The multiplication Ao-A2 is conducted in this period. The multiplication of y-x'4 is not conducted only when the code-length correction is to be executed. In such a case, the creation of error pattern is failed with respect to the code word which requires code-length correction. This, however, does not cause any substantial problem if this code word is set in the parity portion.
- It is thus possible to simplify the encoding and decoding circuit by adopting the circuit arrangement as shown in Fig. 48.
- The syndrome creating circuit, constant output circuit and the pattern output circuit are the same as those described before, so that detailed description thereof is omitted. A description, therefore, will be started again with respect to the K-creating circuit, comparator circuit, and A-creating circuit.
- The K-creating
circuit 481 is a circuit for multiplying the syndrome Si' after code-length correction of formula (34) by α-i for a time corresponding to the code length, thus executing the operation of formula (35). This circuit, therefore, relates only to decoding operation and may be realized by a circuit arrangement similar to the syndrome generating circuit. Practically, however, a circuit as shown in Fig. 49 is used because in this case the computation of EXOR with the received word is not necessary. By using a three-state control switch as the switch shown in Fig. 49 and by adopting a BUS line control construction similar to that of the syndrome generating circuit, a circuit as shown in Fig. 50 is obtained. The syndrome SO does not require code-length correction of the formula (34), but syndromes S1 to S3 have to be converted into S1' to S3' through code-length correction. The code-length correction is conducted by multiplying the respective syndromes with code-length correction constants. When the arrangement is such that corrected syndromes S1' to S3' are output in synchronization with SE, SO is latched at CK3 and is output at KE so as to make the timing of SO coincide with the timing of SO' so that corrected syndromes S', including SO' to S3', are input to the K-creating circuit in the mentioned order. Therefore, the output enable signals RE1 to RE4 of latches for KO to K3 operate in a competing a manner as in the case of the timing chart shown in Fig. 32. These signals RE1 to RE4, however, have to be kept on in the period in which S' is being received by KE and SE. The timing of this operation is shown in Fig. 51. As in the case of the syndrome creating circuit, this circuit has an advantage that it can create K with a small-scale circuit under the supply of the corrected syndromes S' of BUS line structure. In consequence, KO to K3 are output at a period corresponding to four clocks. - Fig. 52 is a block diagram of the comparator circuit. The comparator circuit is intended for a comparison between AO.A1 and A12 so as to judge whether they coincide with each other or not, i.e., whether L2 = A12 + A0, A2 = 0 or
L2 0 0 (formula (39)), and latches the result at CK6, thereby detecting the position of the double-error to be corrected. The position of singular-error is conducted by judging, in accordance with formula (39), whether L1 = A0 = 0 orL1 0 0 and then latching the result at CK6. - The A-creating circuit is composed of a circuit which computes, in accordance with the formula (36), EXOR of KO to K3 created by the K-creating circuit so as to create AO to A2, and a circuit for creating AO + A1, A02 and A12. The A-creating circuit takes part only in the decoding operation. Because of use of a BUS line type construction, the KO to K3 are not input simultaneously. It is therefore, necessary to latch K by MCK by one clock and computing the EXOR of the output delayed by one clock and K, in order to create A. The arrangement of a circuit for realizing such an operation and the timing of operation of this circuit are shown in Figs. 53 and 54, respectively. In Figs. 54 onward, signals shown by hatching are invalid signals. The next portion of A2 is K3 + KO which has no significance in the subsequent algorithms. The output of AO + A1 is conducted only in the portions of A synchronous with A1 and other portions of A does not have any meaning. Signal A is latched at CK7 and CK1 so as to form AO and A1 thereby creating A02, A12. This operation is conducted for attaining as timing optimum for the computation of formulae (37) and (38).
-
- Thus, x2 can be determined by a circuit having a construction shown in Fig. 46, as explained before.
- As will be understood from the foregoing description, this embodiment enables the scale of the circuit to be reduced by virtue of common use of circuit arrangements both by encoding and decoding and multiplied use of circuit, while maintaining the high speed of encoding and decoding operation and while allowing the correction capacity and the code length to be varied.
- In consequence, this embodiment contributes to a reduction in the size and improvement in the performance of any device which incorporates this embodiment of the error correction apparatus of the invention.
- A description will be made hereinunder as to a syndrome creating circuit in a BHC encoding or decoding circuit suitable for use in a fourth embodiment.
- Fig. 55 shows the construction of this embodiment. The circuit shown in Fig. 55 does not incorporate independent circuits for ai but, rather, the scale of the circuit is reduced by making use of the fact that the root αi of the BHC code in formula (2) is progressively carried up from a' (r being any number and set as r = 0 in this case) such as αr+1,αr+1, and so forth. At the same time, a BUS-line structure is used for SO to S3. Since the BUS-line type structure is used for SO to S3, it is necessary that CK and OE (output enable) have to be used in a competing manner in the period corresponding to SO to S3. The CK and OE used for the respective one of SO to S3 are controlled by a signal which is shown in Fig. 56. Signals CKB1, CKB3, CKB5 and CKB7 are signals obtained through inversion of CK1, CK3, CK5 and CK7, by substituting H with L and vice versa. It is therefore necessary that the input J has to be input for each ji in synchronization with CK1. SCL is a signal which takes the L level when the first received word j1 is input. Therefore, SO to S3 are output for every four clocks and the final answer is given in response to each SCL. It is to be understood, however, SPCL is always H in the case of decoding operation.
- The computation of the formula (2) explained above can be applied also to solution of questions in digital FOURIER transformation DFT.
- When it is required that the multiplication by a' is conducted beforehand, an a' circuit is connected between EXOR and the register which outputs J.
- The arrangement also may be such that an αr-ℓ circuit (ℓ represents any positive number) is placed between EXOR and the register which outputs J, while an at circuit is placed between EXOR and the register of SO and the a circuit.
- Fig. 57 illustrates an arrangement which is suitable for use in the case where multiplication by a' is necessary. An aX circuit can be realized by stacking the a circuit shown in Fig. 4 in x-stages.
- Thus, the described embodiment offers the advantages pointed out above, while reducing the scale or size of the circuit, even when the number q of the Galois function GF(2q) is large or when the number i of the syndromes Si is large or when the syndromes Si are connected through BUS line.
- It will be understood that the sizes of devices such as optical disk device, optical card device, opto-magnetic disk device and DAT can be remarkably reduced by using the circuit of this embodiment.
- A fifth embodiment of the invention will be described hereinunder.
-
-
- Therefore, the product Z is created by multiplying the value of y with 1 to α7 and allowing the passage of output yαi when x (i = 0 to 7) is 1, while passing 0 when xi is 0 for calculating Xi• (Y•α1), and then computing EXOR of the outputs Xi• (Y• αi) (i = 0 to 7). The multiplication can be conducted by a circuit having blocks as shown in Fig. 58. Fig. 59 shows the construction of the circuit as marked by o in Fig. 58. In the circuit shown in Fig. 58, the value of y is successively multiplied with a by the a circuit shown in Fig. 4 and known per se, and ANDs of these outputs y•αi and Xi (i = 0 to 7) are computed. The AND outputs are then processed through an EXOR circuit e, whereby the output Z is obtained. In this Figure, a mark shows a BUS line.
- Fig. 60 shows a modification intended for minimizing delay time in the gate circuit which is formed by stacking of a circuits in seven stages as explained before. This modification therefore can increase the maximum processing speed.
- A description will be made hereinunder as to the division circuit.
-
- Computation of a254 can be conducted without difficulty by a ROM. In this embodiment, however, this computation is conducted by making use of a gate circuit. Due to the nature of Galois function, a circuit for computing any Galois function expressed by
-
- Thus, x2 and x4 are determined from x and x6 is created in the period of first clock.
- In the period of the second clock, x'4 is created from x6 and (x4)2.
- In the period of the third clock, X30 is created from X14 and (x8)2.
- In the period of the fourth clock, X 254 is created from X14 and (X30)2.
- If the requirement is only to determine X 254, the above-described multiplications are enough. In this case, however, it is necessary that y• X254 also is to be computed in the period of the fourth clock. To this end, X14 •y2 is used in place of X14 used in the fourth clock. It is assumed that X14• y is computed by another multiplication circuit.
- Fig. 61 shows a block diagram of a circuit which is designed for conducting the above-described series of operations. This circuit is composed of x2, x4 and x8
circuits multiplier 4 andselections circuits -
- Fig. 62 is a time chart illustrating the operation of the circuit shown in Fig. 61.
- Fig. 66 shows a division circuit having different forms of the x2, x4 and x8 circuits 61, 62 and 63, which improve the processing speed of the division circuit considerably.
- A description will be made as to an exponent/vector conversion circuit. When the Galois function is the one which is represented by GF(28), the exponent n is input to the exponent/vector conversion circuit in a binary form (N7 to N0: Ni = 1 or 0) and is expressed as follows. The exponent n is then decomposed for the purpose of creation of an.
- Thus, the exponent/vector conversion can be accomplished by constructing a circuit which outputs a2i when Ni (i = 0, ..., 7) is 1, and successively multiplying the outputs of such a circuit. An example of such a circuit is shown by block diagram in Fig. 67, while the timing of operation of this circuit is shown in Fig. 68. When binary expressions of a code length n are successively fed starting with NO and ending with N7, outputs αNO to α N7•128 are successively obtained at the BUS line Y and are sent to a
multiplication circuit 672. The output Z of themultiplication circuit 672 successively produces values multiplied with αNO to αN7•128 . These outputs are registered in aregister 3 and are input to another input terminal x of themultiplication circuit 672. It is to be noted that the content of the register is beforehand set such that 1 is output to the terminal X when the Y initially outputs α NO 'SO that an is produced when α N7•128 is output. - The construction of the exponent/
vector conversion circuit 671, which is adapted to successively output αNO to α N7•128 in accordance with binary-coded exponents NO to N7, is shown in Fig. 69. - Hitherto, this circuit has been constituted by a ROM. In contrast, the described embodiment realizes this circuit by combination of gates. In Fig. 69, a symbol represents an OR circuit.
- When the irreducible polynominant is expressed by P(x) = x8 + x4 + x3 + x2 + 1 and on condition of Ni = 1 (i = 0 - 7), the circuit shown in Fig. 69 successively outputs a = (01000000), a2 = (00100000), a4 = (00001000), a8 = (10111000), α16 = (00110010), a32 = (10111001), a64 = (11111010) and α128 = (10100001), and outputs 1 on condition of Ni = 0.
- It is thus possible to conduct conversion of exponentially expressed value n into vector expression an by a simplified gate circuit arrangement, in contrast to conventional art in which this conversion is conducted by a ROM.
- The circuit shown in Fig. 67
necessitates 8 clocks for serially outputting NO to N7 till an is formed. When it is desired to reduce the number of clock inputs, the arrangement may be modified such that NO to N7 are output in a parallel form and a plurality of multiplication circuits are used to reduce the number of input clocks. In such a modification, however, the size of the circuit will be slightly increased as compared with that of the described embodiment.
Claims (16)
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP96200874A EP0723342B1 (en) | 1986-09-30 | 1987-09-29 | Error correction apparatus |
EP93201798A EP0566215B1 (en) | 1986-09-30 | 1987-09-29 | Error correction apparatus |
Applications Claiming Priority (14)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP61232007A JP2547744B2 (en) | 1986-09-30 | 1986-09-30 | Encoding / decoding circuit |
JP61232008A JP2566929B2 (en) | 1986-09-30 | 1986-09-30 | Encoding / decoding circuit |
JP232001/86 | 1986-09-30 | ||
JP232007/86 | 1986-09-30 | ||
JP61232005A JPS6386926A (en) | 1986-09-30 | 1986-09-30 | Galois body dividing circuit |
JP61232001A JP2541938B2 (en) | 1986-09-30 | 1986-09-30 | Syndrome generation circuit |
JP61232003A JPH0783277B2 (en) | 1986-09-30 | 1986-09-30 | Original representation format conversion circuit on Galois field |
JP232003/86 | 1986-09-30 | ||
JP232005/86 | 1986-09-30 | ||
JP232006/86 | 1986-09-30 | ||
JP232004/86 | 1986-09-30 | ||
JP61232006A JP2823158B2 (en) | 1986-09-30 | 1986-09-30 | Error correction device |
JP61232004A JPS6386925A (en) | 1986-09-30 | 1986-09-30 | Galois body multiplying circuit |
JP232008/86 | 1986-09-30 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP93201798.1 Division-Into | 1987-09-29 |
Publications (3)
Publication Number | Publication Date |
---|---|
EP0262944A2 EP0262944A2 (en) | 1988-04-06 |
EP0262944A3 EP0262944A3 (en) | 1989-01-25 |
EP0262944B1 true EP0262944B1 (en) | 1994-03-09 |
Family
ID=27566647
Family Applications (3)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP87308648A Expired - Lifetime EP0262944B1 (en) | 1986-09-30 | 1987-09-29 | Error correction apparatus |
EP93201798A Expired - Lifetime EP0566215B1 (en) | 1986-09-30 | 1987-09-29 | Error correction apparatus |
EP96200874A Expired - Lifetime EP0723342B1 (en) | 1986-09-30 | 1987-09-29 | Error correction apparatus |
Family Applications After (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP93201798A Expired - Lifetime EP0566215B1 (en) | 1986-09-30 | 1987-09-29 | Error correction apparatus |
EP96200874A Expired - Lifetime EP0723342B1 (en) | 1986-09-30 | 1987-09-29 | Error correction apparatus |
Country Status (3)
Country | Link |
---|---|
US (2) | US5590138A (en) |
EP (3) | EP0262944B1 (en) |
DE (3) | DE3789266T2 (en) |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3137119B2 (en) * | 1989-06-07 | 2001-02-19 | キヤノン株式会社 | Error correction device |
FR2759218B1 (en) * | 1997-01-31 | 1999-05-07 | Canon Kk | DEVICE AND METHOD FOR PROCESSING INFORMATION SYMBOLS |
US6098192A (en) * | 1997-09-17 | 2000-08-01 | Cirrus Logic, Inc. | Cost reduced finite field processor for error correction in computer storage devices |
JPH11196006A (en) | 1997-12-26 | 1999-07-21 | Nec Corp | Parallel processing syndrome calculation circuit and reed solomon decoding circuit |
FR2785743A1 (en) * | 1998-11-09 | 2000-05-12 | Canon Kk | DEVICE AND METHOD FOR ADAPTING TURBOCODERS AND DECODERS ASSOCIATED WITH VARIABLE LENGTH SEQUENCES |
AU768082B2 (en) * | 2000-02-10 | 2003-12-04 | Sony Corporation | Method for recording and/or reproducing data on/from recording/recorded medium, reproducing apparatus, recording medium, method for recognizing recording/recorded medium, and method for recording and/or reproducing data for apparatus using recording/recorded medium |
US6779014B1 (en) * | 2001-02-15 | 2004-08-17 | Chung-Shan Institute Of Science & Technology | Cyclic step by step decoding method used in discrete fourier transform cyclic code of a communication system |
JP2010049780A (en) * | 2008-07-25 | 2010-03-04 | Panasonic Corp | Ecc circuit, semiconductor storage device, and memory system |
CN102314330B (en) * | 2011-09-09 | 2013-12-25 | 华南理工大学 | Composite finite field multiplier |
TWI457751B (en) * | 2012-07-13 | 2014-10-21 | Univ Feng Chia | Tandem fault tolerant device |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4216531A (en) * | 1978-11-17 | 1980-08-05 | Control Data Corporation | Finite field multiplier |
EP0080528A1 (en) * | 1981-11-30 | 1983-06-08 | Omnet Associates | Computational method and apparatus for finite field arithmetic |
EP0096163A2 (en) * | 1982-06-15 | 1983-12-21 | Kabushiki Kaisha Toshiba | Apparatus for dividing the elements of a Galois field |
EP0152702A2 (en) * | 1983-12-30 | 1985-08-28 | Sony Corporation | Arithmetic circuit of finite field |
Family Cites Families (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4162480A (en) * | 1977-01-28 | 1979-07-24 | Cyclotomics, Inc. | Galois field computer |
US4218582A (en) * | 1977-10-06 | 1980-08-19 | The Board Of Trustees Of The Leland Stanford Junior University | Public key cryptographic apparatus and method |
US4151510A (en) * | 1978-04-27 | 1979-04-24 | Honeywell Information Systems | Method and apparatus for an efficient error detection and correction system |
JPS5724143A (en) * | 1980-07-18 | 1982-02-08 | Sony Corp | Error correcting method |
CA1170776A (en) * | 1980-07-18 | 1984-07-10 | Yoichiro Sako | Method of error correction of blocks of data |
US4413339A (en) * | 1981-06-24 | 1983-11-01 | Digital Equipment Corporation | Multiple error detecting and correcting system employing Reed-Solomon codes |
US4438512A (en) * | 1981-09-08 | 1984-03-20 | International Business Machines Corporation | Method and apparatus for verifying storage apparatus addressing |
NL8104342A (en) * | 1981-09-21 | 1983-04-18 | Philips Nv | CALCULATOR SYSTEM, BASED ON A SYMBOL-CORRECTING CODE WITH TWO WORKING MODES. |
JPS58125175A (en) * | 1982-01-21 | 1983-07-26 | Sony Corp | Multiplying circuit of galois field |
EP0085130A1 (en) * | 1982-02-02 | 1983-08-10 | Omnet Associates | Method and apparatus for maintaining the privacy of digital messages conveyed by public transmission |
JPS58147807A (en) * | 1982-02-26 | 1983-09-02 | Toshiba Corp | Error correcting circuit |
JPS58219851A (en) * | 1982-06-15 | 1983-12-21 | Toshiba Corp | Correcting circuit of error |
EP0096165B1 (en) * | 1982-06-15 | 1988-06-08 | Kabushiki Kaisha Toshiba | Apparatus for dividing the elements of a galois field |
JPS58219852A (en) * | 1982-06-15 | 1983-12-21 | Toshiba Corp | Correcting circuit of error |
DE3483375D1 (en) * | 1983-06-22 | 1990-11-15 | Hitachi Ltd | METHOD AND SYSTEM FOR ERROR CORRECTION. |
US4759063A (en) * | 1983-08-22 | 1988-07-19 | Chaum David L | Blind signature systems |
US4584686A (en) * | 1983-12-22 | 1986-04-22 | Optical Storage International | Reed-Solomon error correction apparatus |
JPH0812612B2 (en) * | 1983-10-31 | 1996-02-07 | 株式会社日立製作所 | Error correction method and apparatus |
NL8400629A (en) * | 1984-02-29 | 1985-09-16 | Philips Nv | QUICK DECODE FOR REED-SOLOMON CODES, WHICH CAN ALSO BE USED AS ENCODEUR AND RECORDING / REPRODUCTION DEVICE WITH SUCH AN ENCODE / DECODEUR. |
US4649541A (en) * | 1984-11-21 | 1987-03-10 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Reed-Solomon decoder |
US4658094A (en) * | 1985-03-28 | 1987-04-14 | Itt Corporation | Encryption apparatus and methods for raising a large unsigned integer to a large unsigned integer power modulo a large unsigned integer |
US4797848A (en) * | 1986-04-18 | 1989-01-10 | Hughes Aircraft Company | Pipelined bit-serial Galois Field multiplier |
JPH06221137A (en) * | 1993-01-27 | 1994-08-09 | Isuzu Ceramics Kenkyusho:Kk | Exhaust gas disposing device |
-
1987
- 1987-09-29 EP EP87308648A patent/EP0262944B1/en not_active Expired - Lifetime
- 1987-09-29 DE DE3789266T patent/DE3789266T2/en not_active Expired - Lifetime
- 1987-09-29 EP EP93201798A patent/EP0566215B1/en not_active Expired - Lifetime
- 1987-09-29 DE DE3751958T patent/DE3751958T2/en not_active Expired - Lifetime
- 1987-09-29 EP EP96200874A patent/EP0723342B1/en not_active Expired - Lifetime
- 1987-09-29 DE DE3752367T patent/DE3752367T2/en not_active Expired - Lifetime
-
1995
- 1995-03-07 US US08/400,521 patent/US5590138A/en not_active Expired - Lifetime
-
1996
- 1996-08-23 US US08/701,327 patent/US5774389A/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4216531A (en) * | 1978-11-17 | 1980-08-05 | Control Data Corporation | Finite field multiplier |
EP0080528A1 (en) * | 1981-11-30 | 1983-06-08 | Omnet Associates | Computational method and apparatus for finite field arithmetic |
EP0096163A2 (en) * | 1982-06-15 | 1983-12-21 | Kabushiki Kaisha Toshiba | Apparatus for dividing the elements of a Galois field |
EP0152702A2 (en) * | 1983-12-30 | 1985-08-28 | Sony Corporation | Arithmetic circuit of finite field |
Non-Patent Citations (6)
Title |
---|
ICASSP 86 Proceedings, April 7-11 1986, Tokyo Japan, vol.2 of 4, pages 22.11.1-22.11.4 R Redinbo, "Fault-tolerant Digital Filtering Structures for Wafer Scale VLSI" * |
IEEE Journal on selected areas in Communications, SAC-4 (1986) Jan no.1, New York, P.A.Scott et al: "A fast VLSI multiplier for GF(2m)", pages 62-66 * |
IEEE transactions on computers, C-33 (1984) October no.10 New York, In-Sek Hsu et al: "The VLSI implementation of a Reed-Solomon encoder using Berlekamp's bit-serial multiplier algorithm" , pages 906-911 * |
IEEE Transactions on Computers, C-34 (1985) August, no.8, New York C.C.Wang et al: "VLSI Architectures for computing Multiplications and Inverses in GF(2m)", pages 709-717 * |
NTIS Technical Notes, 1986, July, no.7, Part B, Springfield Virginia USA C.C.Wang et al: "Multiplier Architecture for coding circuits", pages 792-793 * |
Patent Abstract of Japan, vol.7, no.238 (P-231)(1383), October 22, 1983 & JP-A-58125175 * |
Also Published As
Publication number | Publication date |
---|---|
EP0566215B1 (en) | 1996-11-20 |
US5774389A (en) | 1998-06-30 |
US5590138A (en) | 1996-12-31 |
DE3789266T2 (en) | 1994-09-08 |
EP0723342A3 (en) | 1997-08-20 |
DE3752367D1 (en) | 2003-06-05 |
EP0262944A3 (en) | 1989-01-25 |
EP0723342A2 (en) | 1996-07-24 |
EP0262944A2 (en) | 1988-04-06 |
DE3751958D1 (en) | 1997-01-02 |
EP0566215A3 (en) | 1994-05-18 |
EP0723342B1 (en) | 2003-05-02 |
DE3752367T2 (en) | 2004-02-19 |
DE3789266D1 (en) | 1994-04-14 |
DE3751958T2 (en) | 1997-04-10 |
EP0566215A2 (en) | 1993-10-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP0158510B1 (en) | Error detection and correction in digital communication systems | |
EP0154538B1 (en) | Parity and syndrome generation for error and correction in digital communication systems | |
EP0278383B1 (en) | Error correction method using reed-solomon code | |
US5440570A (en) | Real-time binary BCH decoder | |
US5343481A (en) | BCH error-location polynomial decoder | |
EP0387924A2 (en) | Method and apparatus for decoding error correction code | |
EP0233075B1 (en) | Method and apparatus for generating error detection check bytes for a data record | |
EP0112988A2 (en) | Syndrome processing for multibyte error correcting systems | |
US4312069A (en) | Serial encoding-decoding for cyclic block codes | |
KR890004677B1 (en) | Decoding method and apparatus for cyclic codes | |
EP0262944B1 (en) | Error correction apparatus | |
US4527269A (en) | Encoder verifier | |
US4994993A (en) | System for detecting and correcting errors generated by arithmetic logic units | |
JPH0728227B2 (en) | Decoding device for BCH code | |
EP0393080B1 (en) | Hypersystolic reed-solomon encoder | |
JPH0575479A (en) | Circuit apparatus for identifying, revising and correcting error in data flow | |
EP0781471B1 (en) | Crc/edc checker system | |
US5471485A (en) | Reed-solomon decoder using discrete time delay in power sum computation | |
US6374384B1 (en) | Reed Solomon error correcting circuit and method and device for Euclidean mutual division | |
US6880121B2 (en) | Parallel processing syndrome calculating circuit and reed-solomon decoding circuit | |
EP0004718A1 (en) | Method of and apparatus for decoding shortened cyclic block codes | |
EP0329775A1 (en) | High bandwidth reed-solomon encoding, decoding and error correcting circuit | |
US6859905B2 (en) | Parallel processing Reed-Solomon encoding circuit and method | |
EP0341851A2 (en) | Method and apparatus for interleaved encoding | |
US20230091457A1 (en) | Byte error correction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
AK | Designated contracting states |
Kind code of ref document: A2 Designated state(s): DE GB |
|
PUAL | Search report despatched |
Free format text: ORIGINAL CODE: 0009013 |
|
AK | Designated contracting states |
Kind code of ref document: A3 Designated state(s): DE GB |
|
17P | Request for examination filed |
Effective date: 19890616 |
|
17Q | First examination report despatched |
Effective date: 19900730 |
|
GRAA | (expected) grant |
Free format text: ORIGINAL CODE: 0009210 |
|
AK | Designated contracting states |
Kind code of ref document: B1 Designated state(s): DE GB |
|
XX | Miscellaneous (additional remarks) |
Free format text: TEILANMELDUNG 93201798.1 EINGEREICHT AM 29/09/87. |
|
REF | Corresponds to: |
Ref document number: 3789266 Country of ref document: DE Date of ref document: 19940414 |
|
PLBE | No opposition filed within time limit |
Free format text: ORIGINAL CODE: 0009261 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT |
|
26N | No opposition filed | ||
REG | Reference to a national code |
Ref country code: GB Ref legal event code: IF02 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: GB Payment date: 20060918 Year of fee payment: 20 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: DE Payment date: 20061122 Year of fee payment: 20 |
|
REG | Reference to a national code |
Ref country code: GB Ref legal event code: PE20 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: GB Free format text: LAPSE BECAUSE OF EXPIRATION OF PROTECTION Effective date: 20070928 |